STATUS:
OOF (ONAP Optimization Framework):
1. High Level Component Definition and Architectural Relationships
draw.io Diagram | ||||||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
2. API definitions
OOF provides the following interfaces:
Interface Name | Interface Definition | Interface Capabilities |
---|---|---|
OOFE-1 |
| It enables placement based on a wide variety of policy constraints including capacity, location, platform capabilities, and other service specific constraints. |
OOFE-2 | PCI/ANR Optimization | Enables PCI/ANR optimization API for SON. |
OOFE-3 | Schedule Optimization | a policy driven workflow schedule optimizer for change management planning. This interface schedule workflows in time to maximize parallel change management activities, while respecting dependency between the workflows. |
OOFE-4 | Route Optimization | Provides an interface for Router Optimization. |
OOFE-5 | OOF Model Administrator | This is for the OOF Model Administrator API. This API is a way to dynamically change the optimization models that will be used to find solutions for various optimization problems. This API will be used to Create, Update or Delete Models. |
OOFE-6 | This interface enables slice selection recommendations. |
Note: xxxI interface is a Component internal interface. xxxxE interface is a component external interface
The current API documents can be found at:
The OOF user guide can be found at: OOF latest user guide
- OOF internal APIs can be found: OOF API specification
OOF consumes the following Interfaces:
Interface Name | Purpose Reason For Use |
---|---|
SDCE-1 | For PCI/ANR optimization, OOF Retrieves the Cellsite inventory details from the configdb API, which is hosted as part of the SDNC/R component |
AAIE-1 | OOF interfaces with AAI to retrieve the inventory |
MCE-5 | OOF queries Multicloud for real-time available capacity information |
MUSICI-1 | OOF interfaces with MUSIC to persist service state. |
3. Component Description:
ONAP Optimization Framework (OOF) is an Umbrella project, with the primary goal of addressing the optimization needs of ONAP. OOF is a framework that supports creating and running a suite of Optimizing applications including:
-Change Management Scheduling optimizer
-Homing/Placement optimizer
-PCI optimizer
-Route optimizer
OOF is targeted to be an optimization platform with the following goals:
-Declarative, policy-driven approach to solving optimization problems
-Reusable components: data and policy adapters/libraries, execution environment
-Support General-Purpose as well as Custom optimizers
-Extensible to multiple optimization problems
Internally, OOF has the following components:
- OSDF: Optimization Service Design Framework, which is a collection of APIs and libraries, along with a generic runtime optimizer
- Status: Part of ONAP since Beijing Release.
- HAS: Homing and Allocation Service, which provides a policy based constraint driven selection optimizer
- Status: Part of ONAP since Beijing Release.
- CMSO: Change Management Schedule Optimizer, providing schedule optimization
- Status: Seed code up-streamed in R3
- FGPS: Fine Grained Placement Service
- Status: Seed code up-streamed, POC in Frankfurt
4. Known system limitations
Please find the known system limitations in the following links
5. Used Models
OOF doesn't directly create models, but indirectly consumes them via AAI, Policy and Multi Cloud.
- Service and Resource Info, from: AAI
- Network Topology for CM: AAI
- HPA Flavors/Capabilities/CapacityInfo, from : AAI
- Policy Models (homing, PCI) from: Policy
- Infrastructure Metrics Info (capacity), from: MultiCloud
- Cloud agnostic Intent Info, from: MultiCloud
- AZ level capacity Info, from: MultiCloud (for F-GPS)
- PCI configuration data(not yet a part of SDC model)
- Slice/Subnet Profile and Slice/Subnet Instance models, from AAI
6. System Deployment Architecture
7. New Capabilities in this Release
- Component Upgrades to new Policy Lifecycle API
- Support for E2E Network Slicing
- Support to perform CRUD operations on certain optimization models during runtime.